jru at kth.se
Mon Dec 5 17:00:09 CET 2011
Many thanks for explaining that the right choice of RmtKmax guarantees
that the harmonic and Fourier parts of potential are continuous. I have
consulted Cotteneier's Notes in Wien2k-Textbooks and gone up to
RmtKmax=8.5 for Sr3Mn2O7 using Rmt's from setrmt_lapw.
The result is better than I could believe in my wildest imagination:
vcoul(LM=0,0) minus vcoulPoisson(LM=0,0) as derived from clmsum(LM=0,0)
is constant with respect to radius within 1 mRy.
This is THE excellent self-test for my particular application of Wien2k.
With best regards, John Rundgren
On Tue, 2011-11-29 at 07:55 -0600, Laurence Marks wrote:
> What I would do is use lapw3 to generate the x-ray structure factors
> then convert using the Mott formula and do a FFT (or use the structure
> factors as they are). This is right. We have used this in the past for
> TED/HREM data, and personally I trust this a bit more than the Poisson
> equation solution, and you can easily include an isotropic temperature
> factor for all atoms. (Anisotropic would need an average over
> positions (e.g. look at Physical Review B61 (2000) 2506-2512 and
> Sergei Dudarev's papers at
> http://www.materials.ox.ac.uk/peoplepages/dudarev.html ).
> N.B., you need to ensure that you do not have discontinuities at the
> RMT, so larger LM's than standard and perhaps a higher RKMAX than 7
> are advisable. You can use RMTCheck on the contributed software page
> to check.
> 2011/11/29 John Rundgren <jru at kth.se>:
> > Dear Wien2k Team,
> > This is about vcoul(L,M=0,0) generated by Fe3O4 and Sr3Mn2O7. I compare
> > two potential curves as functions of radius R:
> > (1) VC(R)=vcoul00(R)/sqrt(4*pi), spherically symmetric;
> > (2) VCpois(R), spherically symmetric, obtained from clmsum00 and
> > Poisson's equation (using subroutine charg2.f). My boundary condition of
> > PE is the conventional one giving zero VCpois for a neutral muffin-tin
> > and large R.
> > In summary, I expect that VC and VCpois are equal up to a constant,
> > individual for individual atoms depending on the SCF boundary condition
> > in Wien2k. Results for Fe3O4 and Sr3Mn2O7:
> > Fe3O4, Table for potential value at RMT:
> > Fe1 Fe2 O
> > VC -0.13 0.18 0.37
> > VCpois -2.03 -2.08 -1.15
> > VCpois(R)=VC(R)+const with great accuracy.
> > Sr3Mn2O7 with RMT's from setrmt_lapw, Table for potential value at RMT:
> > Sr1 Sr2 Mn O1 O2 O3
> > VC -472 -382 -231 -258 -230 -250
> > VCpois -2.0 -2.0 -2.6 -7.3 -7.5 -7.0
> > First observation: the values differ by factors 35-235. How is the
> > boundary condition for vcoul defined in Wien2k?
> > Second observation: dVC/dR is <0 and dVCpois/dR is >0 for R in the
> > neighborhood of RMT. Indeed, the latter behavior corresponds to a
> > muffin-tin with negative charge.
> > Third observation: VCpois(R) .neq. VC(R)+const. The disagreement is
> > significantly great, see 1st Attachment.
> > When the Fe3O4 calculation is used for LEED, the agreement
> > theory-experiment is satisfactory [Surf.Sci. 602(2008)1299]. On the
> > other hand, a LEED investigation on Sr3Mn2O7 would respond badly on a
> > potential disagreement like the one demonstrated in the 1st Attachment.
> > 2nd Attachment is SR3Mn2O7.struct.
> > VC versus VCpois is a dilemma for Sr3Mn2O7. Is there an expert's input
> > to w2web that would make the potentials equal? A discussion would be
> > much appreciated.
> > With best regards,
> > John Rundgren, KTH, Stockholm
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
More information about the Wien